Tokyo, February 10, 2023 - NEC Corporation (NEC; TSE: 6701) and the University of Tsukuba in Japan have developed a first-of-its-kind technology (*1) that uses AI to detect edema (swelling) in facial images, which may contribute to the early detection of illness.

Edema can be caused by a wide range of issues, including conditions with the kidneys, heart and liver. Technologies that routinely check the status of edema can help to identify changes in a person's condition, which can lead to the early detection and treatment of an illness before it becomes more serious.

Conventionally, dialyses patients, for example, have used scales to measure body weight and estimate edema. This new technology, however, uses facial images and AI to estimate the degree of an individual's edema. An experiment using data from 39 dialysis patients confirmed the presence of edema with an 85% accuracy, and the degree of the edema could be estimated with a 0.5 kg mean absolute error for body weight changes (*2). This indicates that the new technology can be a reliable alternative to traditional weight measurement.

Facial changes due to edema are often minimal and vary significantly from person to person. This makes it necessary to learn and construct an estimation model of facial changes for each individual who uses the system. However, collecting large amounts of data from a single patient, for example, can be a large burden.

With this new technology, NEC and the University of Tsukuba used facial images from multiple patients to pre-train an AI model that extracts information on various types of edema that appear on the face. In doing so, the organizations developed a method for pre-learning the presence and degree of edema with high accuracy by using body weight, which is correlated with edema, as training data. By using this pre-trained AI model as a base, even with a small amount of patient data, it is possible to transfer the pre-trained AI model to a model tailored to each patient's edema and improve estimation accuracy.

NEC's face recognition software (*3) is used by this technology to detect faces rapidly and accurately from images that can be taken with smartphones or tablets, enabling facial data to be obtained quickly and easily regardless of the location or environment. Using these images, AI-driven models are able to regularly analyze the changes in a patient's face, which may contribute to the early detection and treatment of an illness.

Going forward, NEC and the University of Tsukuba will continue to collaborate on the accumulation of data for further improving this technology and exploring specific applications in the healthcare field. NEC aims to see the practical application of this technology in fiscal 2024.

This research was conducted under a joint research agreement between NEC Corporation and the University of Tsukuba.

***

Notes:

About NEC Corporation
NEC Corporation has established itself as a leader in the integration of IT and network technologies while promoting the brand statement of “Orchestrating a brighter world.” NEC enables businesses and communities to adapt to rapid changes taking place in both society and the market as it provides for the social values of safety, security, fairness and efficiency to promote a more sustainable world where everyone has the chance to reach their full potential. For more information, visit NEC at https://www.nec.com.